Child support awards : differentials and trends by race and marital status

UNIVERSITY OF

ILLINOIS LIBRARY

AT URBANA-CHAMPAIGN

BOOKSTACKS

CENTRAL CIRCULATION BOOKSTACKS

The person charging this material is re-
sponsible for its renewal or its return to
the library from which it was borrowed
on or before the Latest Date stamped
below. You may be charged a minimum
fee of $75.00 for each lost book.

Theft, mutilation, and underlining of books are reasons
for disciplinary action and may result in dismissal from
the University.

TO RENEW CALL TELEPHONE CENTER, 333-8400
UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN

JAM 2 1 1997

L)LU 0 4 2000

When renewing by phone, write new due date below
previous due date. L162

>4 COPY 2

t53&

LTY WORK]
P&PER NO. 1124

Child Support Awards: Differentials and
Trends by Race and Mental Status

THE LIBRARY OF THE

Andrea H, Seller
John W. Graham

UNIVERSITY OF ILLINOIS

college

■versify o? :

- c- iLi Business Adn'intstrai
ano 6- !':jln^33 Research

FACULTY WORKING PAPER NO. 1124
College of Commerce and Business Administration

University of Illinois at Urbana-Champaign

March, 1985

Child Support Awards: Differentials and Trends
by Race and Marital Status

Andrea H. Seller, Assistant Professor
Department of Family and Consumer Economics

John W. Graham, Assistant Professor
Department of Economics

This research was supported in part by a grant from the National Institute
of Child Health and Human Development #1 R01 HD19350-01. we wish to
acknowledge the excellent computer programming assistance of John Boyd
and research assistance of Kee-ok Kim Han and Edwin Sexton.

Digitized by the Internet Archive

in 2011 with funding from

University of Illinois Urbana-Champaign

http://www.archive.org/details/childsupportawar1124bell

AESTRACT

Child Support Awaras: Differentials and Trends

by Race and Marital Status

The purpose of this paper is to examine differentials in the
award of child support by race and marital status. It also
evaluates the impact of the child support enforcement program,
begun in 1975, on the incidence of child support awards. The
analysis is based upon data from the 1979 and 1982 March/April
Match Files of the Current Population Survey. According to
these data, never-married mothers are significantly less likely
than ever-married mothers to have a child support award, but
which of the never-married do cannot readily be explained by our
model. Ever-disrupted black marriages were considerably less
likely than nonblack marriages to result in a child support
award, but this has been changing, especially since the passage
of the child support enforcement program in 1975. Finally, we
find that the value of neither new nor old awards is keeping up
with the cost of living or with the increase in men's earnings.

Child Support Awards: Differentials ano Trends
by Race and Marital Status

Severe economic consequences are resulting from the rapid rise in
single-parent families headed by women. These families typically have fewer
resources and more restricted opportunities than two-parent families, and
often have little choice but to rely upon the welfare system. As a result,
public attention has turned to consider the extent to which absent fathers
neglect to provide for the support of their children, leaving support instead
to the woman and/or the state.

The numbers of single-parent families headed by women are growing
rapidly. By 1981, over 15 percent of a'll families were headed by a woman only
(U.S. Bureau of the Census, 1982, Table 60). The probability that a marriage
will end in divorce is high, and until 1982 had been rising. Increasing rates
cf illegitimacy among the never-married also account for considerable growth
in the numbers of female-headed families. Recent estimates suggest that
two-fifths of children born to married women will experience the disruption of
their parent's marriage and will live for some time with only their mother
(Bumpass, 1984).

The incidence of poverty among single-parent families headed by a
woman is far greater and is growing more rapidly than among other types of
families, causing concern over "the feminization of poverty." In 1980, the
median income of female-headed families was $10,830 compared to $23,180 for
married-couple families (U.S. Bureau of the Census, 1982, Table 717). Of the
6.2 million families with incomes below the poverty level in that year, almost
3 million were female-headed (U.S. Bureau of the Census, 1982, Table 735). Cf
the 7.1 million women with children from an absent father in 1979, about 2
million had incomes below the poverty level (U.S. Bureau of the Census, 1981).

-2-
AFDC payments as well as a large share of benefits from other public aid
programs go to support these families.

Economic well-being of these families is higher if they are awarded
child support. The poverty rate in 1978 for women awarded child support was
18 percent, while for those not awarded support it was 42 percent (U.S. Bureau
of the Census, 1981). Unfortunately, many eligible families do not have a
child support award; in 1979 only 59 percent of mothers with children present
from an aosent father had an award. Nonblack mothers were more likely to have
an award than black mothers, and ever-married mothers were more likely than
never-married mothers. Among nonblack mothers 70 percent haa an award, while
among black mothers only 29 percent did. Around 78 percent of ever-divorced
mothers and 45 percent of separated mothers had an award, but only 11 percent
of never-married mothers did (U.S. Bureau of the Census, 1981). While the
award rate remained unchanged between 1979 and 1982 for all mothers, it rose
for black mothers from 29 to 34 percent and for never-married mothers from 11
to 14 percent (U.S. Bureau of the Census, 1983).

The increase in the rate of child support awards among the black ana
never-married populations may be due to the success of recent efforts in the
area of chile support enforcement. In 1975, as Title IV-D of the Social
Security Act, Ccngress passed a program to aid in the collection of child
support from absent parents and thereby to reduce the burden en the welfare
system. Unaer the IV-D program, state agencies engage in activities to locate
absent parents, to establish paternity, to establish child support
obligations, and to enforce such obligations. Since its inception the program
has grown considerably; thus, the increases observed in the rate of child
support awards among blacks and never-married mothers may possibly be the
result of federal and state efforts in this area.

-3-

The primary purpose of this paper is to examine these differentials
in the award of child support between black and nonblack and between
never-marriea ana ever-married mothers. One aspect of this involves
explaining the upward trend over time in the likelihood of having a child
support award among blacks. A second goal of this paper is to evaluate the
impact of the child support enforcement program, begun in 1975, on the
incidence of child support awaras. And finally, for ever-married mothers
only, we analyze factors affecting the amount of their child support awards
and trends over time in these amounts, assessing the extent to which they keep
up with the cost of living.

Empirical studies on the determinants of child support awards have
been undertaken only recently; thus, relatively little conclusive evidence is
available. Gaps in our knowledge include racial and marital status
differences and the effects of legislation. Beller and Graham (1985) and
Cassetty (1978) both found evidence to suggest significant racial differences
in the oeterminants of child support. Moreover, previous studies of the
economic well-being of female-headed families (Ross and Sawhill, 1975;
Bradbury et al. , 1979) emphasize the importance of race as a factor. We know
little about whether the oeterminants of child support differ for
never-marriea mothers, largely black and on AFDC, from those for ever-married
mothers. Significant differences between the results of Sorenson and
MacDonald (1983) for AFDC mothers and Seller and Graham (1985) for
ever-married mothers suggest that this may be the case. Implications for
whether child supoort enforcement can be uniformly successful in AFDC and
ncn-AFDC populations may aepena upon any such differences.

The source of the Census estimates aiscussed above and the oata sets
upon which our analysis is based are the 1979 and 1982 March/April Match Files

-4-

of the Current Population Survey (CPS). The special April supplements to the
1979 and 1982 CPS were cesigned to collect detailed data from all women 18
years of age and older on their marital status, divorce history, and the award
and receipt of child support and alimony payments. These data were combined
in a file with the income and demographic data from the March CPS. Combining
these two years of data yields a sample of 7680 mothers with own children
under 21 years of age present1 from an absent father, of whom 4004 are
divorced or separated, 2208 are remarried and 1468 are never-married. Out of
all of these families, 1163 live with parents or other relatives. ^ There
are 1868 black families.

As shown in Table 1, there are significant differences in the
percentage of women awarded child support by race and marital status. Among
ail women, blacks are much less likely to have a child support award than
nondacks. Part of the reason is that blacks are more likely to be currently
separated or never-married — marital status groups with lower probabilities of
award than the ever-divorced. Black women comprise around 61 percent of
never-married mothers; furthermore, while black women comprise 35 percent of
currently separated women, they make up only 11 percent of ever-divorced
women. The low proportion of blacks who divorce is reflected in the
considerably longer perioa of separation of black than of white females, 10 as
compared to 2 years (Espenshade, 1983). The numbers in parentheses in Table 1
show that blacks form an increasing proportion of separated mothers, the
longer ago the year of separation. However, this disproportionate membership
of clacks in groups with lower probabilities of award Goes not completely
account for all racial differences in the award of child support. There are
also significant racial differences within some marital status groups. Among
tie ever-divorced, 81.1 percent of nonblack mothers had an award but only

-5-

62.3 percent of black mothers did. Among currently separated mothers, 49.5
percent of nonblacks had an award, while only 33.6 percent of blacks did.
However, among never-married mothers, there is no racial difference in the
probability of an award.

Table 1 about here

Table 1 also shows trends by race in the likelihood of a child
support award at divorce or separation. In general, there has been an
upward trend in the probability of a child support award both among the
separated and the ever-divorced. The upward trend appears more pronounced for
blacks than for nonblacks; among the separated the trend appears especially
strong since 1975.

In this paper we will attempt to explain these racial and marital
status differences in the award of child support and these trends over time.
First, we estimate the determinants of the probability of having a child
support award using maximum likelihood logit for the entire sample and for
ever-married and never-married mothers separately. Then, for ever-married
mothers only (since so few of the never-married are awarGed support), we
estimate the determinants of the amount of cnila support due in the year prior
to the survey (1978 or 1981), using ordinary least squares (OLS). Beth the
probability and the amount of the award are estimated with essentially the
same set of explanatory variables, although by necessity somewhat different
variables appear in the never-married equation. In addition to race, the
explanatory variables include a vector of characteristics for the needs of the
mother and her children (mother's age or age at divorce, education, and the
number of chilcren), proxy variables for the financial ability of the absent

-6-
father to provide support (eGucation and age of the mother), and proxy
variables for the legal environment, to be discussed below. We also control
for current marital status (separatee or remarried) and length of the
separation. A definition of variables is shown in the Appendix, Table A-l.

Variables that increase the mother's neeas or the absent father's
ability to pay shoula increase the likelihood and amount of an award.
Although we do not expect remarriea women to differ from other women in the
award of child support since remarriage occurs after the award, we control for
remarriage in the regression to test this hypothesis. As suggested in Table
1, separated women are less likely than divorced women to have a child support
awaro. Among the currently separated, those separated long ago are less
likely than those separated recently to have an award. This may be due to a
rising trend in the probability of an award or to the fact that women who do
not divorce shortly after (say, within two years of) separation are less
likely ever to obtain a formal support award.

We measure the legal environment at the time of the disruption
through the use of an historical trend line. We allow this trend in the award
of child support to change after the passage of the child support enforcement
program in 1975 with a secona trend line beginning in that year. Since the
program expanded over time, we expect change to be continuous rather than show
one single improvement at the time of the law's passage. This formulation
allows us to separate any historical upward trend in the award of child
support from an upward trend due to Title IV-D. Geographic location measures
should also help to capture regional variation in the award of child support.

-7-
EMPIRICAL RESULTS
Probability of a Child Support Award

All else equal black mothers are less likely to have a child support
awara than ncnblack mothers. However, all else is not equal. Blacks are also
more likely to be young, never-married, to live in central cities and to have
less education than nonblacks — all characteristics which reduce their chances
of being awarded child support. In this section we seek to determine how much
of the large observed racial aifference in award probabilities can be
accounted for by these other factors. In addition we look at changes over
time in the probability of a child support award.

The probability of having a child support award was estimated first
for our entire sample of 7680 mothers. Let P be the probability that child
support is awarded or agreed to. Then the logistic function

P = 1/(1 + e-6X-u)
was estimated by maximum likelihood methods, where X is a vector of
independent variables and B a vector of coefficients to be estimated. Table 2
presents the estimated partial derivatives of P with respect to each variable,
found by multiplying the estimated p's by P(l-P), where P is the mean of the
dependent variable. Asymptotic t-ratios appear in parentheses.

In general, these results are consistent with earlier findings on
the probaoility of award among ever-married mothers (3eller and Graham,
1985). Older and more educatea mothers are more likely to have a child
support award, but the impact of age and education appear nonlinear (EDUC is
positive, but COLLGRAD is negative; AGE is positive, but AGESQ is negative).
The greater the number of children (PATERNR), the more likely there is a child
support award. Compared to women in the West, those in the Northeast (NEAST)
are less likely and those in north central states (NCENTR) are more likely to

-8-

have an award. Women in central cities (CO are less likely to have an award.

Taole 2 about here

Among the strongest determinants of whether or not child support is
awarded are marital status, race, and ethnicity. Ceteris paribus, compared to
divorced mothers, never-married mothers are 73.2 percent less likely to have
an award, separated mothers are 28.5 percent less likely and those separated
for more than two years are an additional 10.3 percent less likely. In this
equation, remarried women appear to be less likely to have an award than
divorced women; however, when we include a time trend in the model (as in
column 3), this effect becomes insignificant. Blacks are 21.1 percent less
likely (in 1979) than nonblacks and women of Spanish origin are 12.0 percent
less likely than non-Hispanics to have an award. For all women together,
there is no evidence that the probability of being awarded child support
differed between the 1979 and 1982 samples (YEAR82 is insignificant), but for
black women there is some evidence of a positive trena over time: BLYEAR82 is
positive and significant, indicating that a randomly selected black woman from
the 1982 CFS is 7.9 percent more likely than her counterpart from the 1979 CPS
to have a child support award.

The large difference between the proportion of ever-marriea and
never-married mothers with a child support awara suggests it may be important
to analyze ceterminants of awards for these two groups separately. Moreover,
since there is no information in the aata on when the never-married secured
their child support award, while there is information on the date of marital
disruption for ever-married mothers, trencs in tne award of child support can
only be analyzed for the latter group. Results for the never-married mothers
appear in column 2 ana for the ever-married in column 3.

-9-

Our findings suggest that the never-married do differ significantly
from the ever-married in the award of child support. The most surprising
differences are that among the never-married, black mothers are no less likely
than nonblack mothers to have an award, while Hispanic mothers are less likely
than non-Hispanic mothers. Moreover, the number of children does not affect
award probability among the never-married. In fact according to these logit
estimates very little matters for the never-married. But, we do observe a
positive time trend in the award of child support for them (YEAR82 is positive
and significant at the 10% level), which might reflect improved methods and
greater federal and state efforts to establish paternity under the child
support enforcement program.

According to the results in column 3, the strongest determinants of
whether or not ever-married mothers are awarded child support when they
experience a marital disruption include race, whether or not the marriage
ended in divorce, and date of the disruption. Among ever-married mothers with
comparable characteristics, blacks were 13.8 percent less likely to have a
child support award (in 1960 or before) than nonblacks. (Racial differences
will be discussed further below.) Women separated two years or less (SEP) are
25.2 percent less likely than ever-divorced women to have secured a child
support award. Women separated for more than two years (SEPGT2) — the average
duration of separation for whites — are an additional 5.9 percent less likely
to have an aw?ra. Among ever-divorced women, as expected, those who have
remarried are neither more nor less likely than the currently divorced to have
an award (REMAR is insignificant). The more recent the marital disruption
(TIME), the more likely that it resulted in a child support award. This
eviaence on trends over time in the probability of child support awards will
oe discussed in detail below.

-10-
Racial Differences

The regression results in Table 2 are consistent with the evidence
in Table 1 that the proportion of olack women with a chilo support award is
well below that of nonblacks and that the size of this racial difference
varies across marital status groups. Among the never-married, there exists
virtually no racial difference in the likelihood of having a child support
award, while among the ever-married (ever-divorced and currently separated)

48.8 percent of blacks have a child support award compared with 75.9 percent
of nonblacks, for a gross Differential of 27.1 percentage points. The
regression coefficient on BLACK in column 3 of Table 2, -13.8, indicates that
almost half of this gross racial differential among the ever-married can be
attributed to differences in economic, demographic and legal factors, while
the other half (13.8/27.1) remains unexplained.5

The question that remains to be answered is why, until at least
fairly recently, ever-disrupted black marriages were so much less likely than
nonolack marriages to result in a child support award. One obvious
explanation is the lower ability to pay of black than of nonblack husbands due
to their lower income and higher unemployment rate. For example, the ratio of
weekly earnings of black men to white men rose from 69.5 percent in 1971 to

71.9 percent in 1981 (Seller and Blau, 1984). However, it is not obvious that
a lower ability to pay should preclude any awara, although it should clearly
recuce the award amount. Moreover, the increase in black male earnings was
small compared to the extent to which the racial gap in the probability of an
award has narrowed.

Another possible explanation is that black women have less incentive
to seeK an awara than nonblack women. If the expected award is small, then
women may have little incentive to incur the costs of getting it. Moreover,

-li-
the perceived need of black mothers may be less than of nonblack mothers if
the former place less stigma on going on welfare than the latter (for a
discussion of welfare stigma, although not as it applies to race, see Moffitt,
1983). Because the relative income of black husbands and wives is closer than
for nonblacks (the earnings of black women as a percent of black men's were
67.6 percent in 1981 as compared to 52.3 percent for whites [Beller and Blau,
1984]), when a marriage is aisrupted, the loss in economic status is less for
black women than for nonblack women and the potential gain from the husband is
less, so the overall incentive to pursue an award is less.

Another set of possible explanations for racial differences in award
center on legal considerations. It may be that the legal system discriminates
against black wives in the award of child support and that recent changes in
the system have reversed that tendency. It is also possible that blacks avoid
using the courts and lawyers or that their ability to use them is less. A
related factor is that blacks tend to have a longer period of separation than
nonblacks. A longer separation may attenuate the bonds between the father and
his child, making an award at divorce less likely. Among other things, the
father may be more difficult to locate at this point. A related consideration
is that in the case of separation the father may in fact be contributing to
the support of his child, but this is not formalized into a legal award, nor
is his continued presence acknowledged, so that his family maintains
eligibility for welfare.
Trends Over Time

To what extent has the probability of securing a child support award
been changing over time? As mentioned above, data limitations force us to
restrict our attention to the ever-married sample in answering this question.
As shown earlier in Table 1, the proportion of ever-married women awarded

-12-
child support has been increasing over time and this upward trend has been
stronger for blacks than for nonblacks.

These general findings are confirmed by the regression results in
Table 2, column 3. The coefficients on TIME and LAW indicate that between
1960 and 1975 the proportion of ever-married nonblack women obtaining a child
support award increased by 1.3 percent per year, but since 1975 the proportion
has actually declined slightly — by 0.4 percent per year. The pattern is
different for black women. For them, the award rate rose more slowly, by 0.8
percent per year, between 1960 and 1975 — although the difference (BLTIME) is
statistically insignificant; however, since 1975, the proportion of black
women awarded support, unlike that of ncnblacks, not only continued to
increase, but the annual rate of increase doubled to 1.6 percent per year.
Thus, since the passage in 1975 of the child support enforcement program, the
ceteris paribus racial difference in the probability of being awarded child
support narrowed significantly. This occurred in part because of a small
aecline in award probabilities among nonblacks and in part because of a large
increase in award prooabilities among blacks.

One goal of the federal child support legislation enacted in 1975
was to increase the proportion of mothers awarded child support. Our results
suggest that at least through 1981 this legislation was successful in
achieving this goal only among blacks. Part of the reason for this may be
that at least initially state IV-D agencies concentrated their efforts on
ootaining and enforcing child support awards for AFDC families — families that
are disproportionately black. According to the Office of Child Support
Enforcement, as of tne close of fiscal year 1979, 85 percent of the 4.9
million open cases of IV-D agencies were AFDC cases. By September 30, 1983,
AFDC cases still representea 78 percent of the 7.5 million open cases (U.S.

-13-
Department of Health and Human Services, 1983, pp. 76-80). While this
emphasis on AFDC cases may explain why, after the passage of Title IV-D, the
trend in the probability of having a child support award among blacks
accelerated, what remains unresolved is why the positive time trend among
nonblacks ceased.
Amount of Child Support Due

The 3542 ever-divorced or currently separated women awarded child
support and due it in either of the sample years (1978 and 1981) are analyzed
together using ordinary least squares to estimate the impact of various
economic and demographic factors on the amount of child support due. The
results appear in Table 3, where the dependent variable is measured as the
amount of child support due in 1978 dollars and coefficients for two almost
identical sets of independent variables appear in the two columns. In
general, the results confirm earlier findings based upon the 1979 data only
(Beller and Graham, 1985): factors which increase the likelihood of an award
also tend to increase the value of the award.

Blacks are awarded significantly less child support than nonblacks.
The mean amount of child support due in constant 1978 dollars was $1832;
ceteris paribus, the average amount due was $354 less if the mother was black,
or about 20 percent less. This may be fully accounted for by the lower income
of black husbands, but, it could also be caused by any of the other factors
discussed above such as differential attitudes toward welfare, differential
use of legal services, and/or discrimination by the courts.

Table 3 about here

Child support awards are not keeping up with the cost of living. In
real aollars, according to the coefficient on YEAR82 in column 1, the average

-14-

amount due was $392 less (about 21 percent less) in 1981 than in 1978, ceteris
paribus. The real amount due women Divorced in 1979 or after (DIVYRGE79 in
column 2) is higher than that due women aivorcea before 1979, but not enough
to make a statistically significant difference.

Awards made seme time ago do not keep up with inflation because they
typically do not have automatic adjustment clauses in them (Krause, 1980), but
what about new awaras? To assess this, we split our samples into individuals
whose marriages were disrupted two years prior to the survey (new
aisruptions), i.e., in either 1977 or 1980, and those whose marriages were
disrupted three or more years prior to the survey (old disruptions), i.e., in
1976 or earlier or in 1979 or earlier. When we compare these old disruptions
from the two samples, we find that for marriages disrupted in 1957-1976 the
average amount due in 1978 in current dollars was $1230 per child, while for
marriages disrupted in 1960-1979 the average amount due in 1981 in current
dollars was $1427 per child. Thus, for old marital disruptions, between 1978
and 1981, the average amount of child support due per child increased 16.0
percent. During this same period, the Consumer Price Index (CPI) rose 39
percent and median male earnings rose 28 percent (U.S. Bureau of the Census,
1982, Table 671). For marriages disrupted in 1977, the average award was
$1473 per chila whereas for those disrupted in I960 the average award was
$1656 per chile. Thus, among new disruptions, the average award amount per
child went up by only 12.4 percent. Surprisingly, the real value of new
awards is decreasing mere than the real value of old awards.

Thus, another important finding of this paper is that the value of
neither new nor old child support awaras is keeping up with the cost of living
or with the increase in men's earnings. This is a surprising finding that has
importance for the increasing immiseration of female-headed families. Even

-15-
women with child support awards are becoming less able to support their
children. An important question for the courts is why their award formulas do
not seem to chance over time along with prices and earnings. Part of the
reason that awards are not keeping up with the cost of living may be that as
the probability of getting an award increases, the average ability to pay of
those with obligations decreases.

SUMMARY AND CONCLUSION
This study has analyzed data from combined 1979 and 1982 April
supplements to the Current Population Survey to study differences in the award
of child support by race and marital status. The following findings emerge
from this study:

1. The percentage of women with children present from an absent father who
are awarded child support varies greatly by race and marital status.
Among all women, nonblacks are more than twice as likely as blacks to have
a child support award and the ever-married are almost six times as likely
as the never-married to have an award. Among the ever -married, currently
separated women are approximately half as likely as the ever-divorced to
have secured an award.

2. The lower probability of child support awards among blacks can be
attributed in part to their disproportionate membership in marital status
groups with lower award probabilities. Blacks are four and one-half times
as likely as nonblacks to be among the never-married and almost twice as
likely to be among the currently separated.

3. Racial differences in award probabilities exist within all marital status
groups except never-married. Among the currently separated, blacks are
one-third less likely than nonblacks to have an award. Among the

-16-
ever-divorced, blacks are almost one-fourth less likely than nonblacks to
have an award.

4. Among the never-marriea, unlike the ever-married, virtually no
statistically-significant socioeconomic characteristics appear to
distinguish mothers who have a child support award from those who do not.

5. Among all women, 50 to 60 percent of the gross racial differential in
award rates can De explained by observed differences in such economic and
demographic characteristics such as marital status, educational
attainment, age, place of residence, and number of children. Among the
ever-married, 50 percent of the gross racial difference can be explained
by these factors.

6. Among the ever-married, the likelihood of being awarded child support at
marital disruption has increased over time, but this upward trend has been
different for blacks and nonblacks. Among nonblacks, the proportion of
women obtaining a child support award increased 1.3 percent per year
between 1960 and 1975 and then declined 0.4 percent per year since then.
Among blacks, the proportion increased 0.8 percent per year between 1960
and 1975 and then accelerated to 1.6 percent per year since then.

7. The overall impact of Title IV-D child support legislation on the
probability of obtaining a child support award has been slight between
1975 and 1982. Only among blacks have award percentages increased. This
can probably be attributed to the fact that most state IV-D agencies
concentrated their resources on AFDC families — families that are
disproportionately black.

8. Neither new nor old child support awarcs are keeping up with the cost of
living or with the increase in men's earnings.

-17-
Several unresolved policy issues are raised by these findings.
While our regression analysis demonstrates that part of the large racial
difference in award rates can be attributeo to differences in socioeconomic
characteristics, a sizable difference remains unexplained. While part of this
may be due to differences in taste, some may also be caused by differential
access to the legal system or discrimination. Another, and perhaps related,
issue raised by our analysis surrounds the dissimilar trend in award rates
experienced by blacks and nonblacks since 1975. If state IV-D agencies begin
to direct more of their efforts towards the non-AFDC population, as recent
trends suggest they might, how will this affect the gains made by blacks? A
final concern raised by our analysis is the declining real value of both new
and old child support awards, an important factor contributing to the
increasing feminization of poverty. Continued inflation requires that old
awards be renegotiated more frequently and that new awards (including court
guidelines that set them) be automatically tied to rising prices or earnings.

-18-
Note

1. Actually, not all of the children are under 21 years of age. Because of
the way the Census asked the questions, if there was anyone present in the
househola under 21, they proceedea with the questions about child
support. Thus, it might be that one child is under 21 and others are over
or that someone else, such as a grancchild or a child from the present
marriage, was under 21 and all children from the absent father were over
21. Mothers were not asked about their children only if there were no
household members under 21 present. This is not very critical for the
discussion of the award of child support, although it would be more so for
its receipt. For further discussion, see Graham and Seller, 1984.

2. Accoroing to an economic analysis, families would choose to live as a
subfamily in the household of a relative in order to share housing costs
and to take advantage of the public goods aspects, the economies of scale,
and the division of labor. These are lost when for example the intact
household divides into two at divorce. The absence cf a chilo support
award may be one reason that female-headed families choose to live as
subfamilies. But, since current living arrangements would clearly not
influence past child support awards, we can thus combine subfamilies with
family heaas in our sample which is especially important for the
never-married group, because 43 percent did not maintain their own
householas.

3. The data do not allow us to determine the year in which a child support
awara was obtained for never-marriea mctners. Thus, we are unable to
examine trends in the award of child support for this group. All we know
is that never-married women in the 1962 sample are more likely to have an
awara than those in the 1979 sample. For women who are currently

-19-

separated or ever-aivorced, we assume that the year of the marital
disruption is the year in which child support was awarded.

4. The reason is that remarried women are likely on the average to have been
divorced longer ago than divorced women, when, as we will see, the
probability of securing a child support award was lower. We cannot
control for time in this equation since it includes the never-married.

5. A similar calculation can be made on the results in column 1 of Table 2
for all women, but it must be done separately for each sample year due to
the inclusion of the variable YEAR82. In the 1979 CPS sample, 69.6
percent of nonblack mothers and 29.3 percent of black mothers had a child
support award, for a gross differential of 40.3 percentage points.' The
regression coefficient on BLACK of 21.1 suggests that among women from the
1979 CPS sample (YEAR82 and BLYEAR82 equal zero), about 48 percent of the
gross racial difference in award probabilities can be explained by
differences in economic and demographic characteristics, while the other
52 percent (21.1/40.3) remains to be accounted for by race. In the 1982
CPS sample, 69.4 percent of nonblacks and 33.5 percent of blacks had a
child support award for a differential of 35.9 percentage points. In our
regression, the relevant racial coefficient for these women is 13.2 (-21.1
on BLACK plus 7.9 on BLYEAR82). This suggests that about 63 percent of
the gross racial difference in award probabilities can be explained by
other factors in the regression, while 37 percent (13.2/35.9) remains to
be accounted for by race.

6. This figure is ootained by adding the coefficients on TIME, BLTIME, LAW
and BLLAW.

7. For old disruptions, the total amount due in 1978 was $1998 for a sample
of 1131; this increased in 1981 to $2212 for a sample of 1318. The

-20-
percent increase was 10.7, smaller than the increase per child because the
number of children per disruption declined. For new disruptions, the
total amount due in 1978 was $2623 for a sample of 193; this increased in
1981 to $2836 for a sample of 254. The percent increase was 7. A.

References
Seller, Andrea H. and Francine Blau. 1984. Trends in earnings differentials

by sex and race: 1971-1981. Paper presented at the American Economic

Association Annual Meeting.
Beller, Anarea H. and John W. Graham. 1985. Variations in the economic

well-being of divorced women and their children. In Martin David and

Timothy Smeeding, eds., Horizontal Equity, Uncertainty, and Economic Well-

Beinq, Studies in Income and Wealth Series, Vol. 50, Cambridge, MA:

National Bureau of Economic Research.
Bradbury, Katherine, Shelaon Danziger, Eugene Smolensky, and Paul Smolensky.

1979. "Public assistance, female headship, and economic well-being."

Journal of Marriage and the Family 41 (August): 519-35.
Bumpass, Larry L. 1984. Children and marital disruption: A replication and

an update. Demography 21 (February): 71-82.
Cassetty, Judith. 1978. Child support and public policy. Lexington, MA:

D. C. Heath and Co.
Espenshade, Thomas. 1983. Black-white differences in marriage, separation,

divorce and remarriage. Paper presented at the Annual Meeting of the

Population Association of America.
Graham, John and Andrea Beller. 1984. A Note on the Number and Living

Arrangements of Women with Children Under 21 from an Absent Father:

Revised Estimates from the April 1979 and 1982 Current Population

Surveys. Unpublished manuscript.
Krause, Harry D. 1981. Child support in America. Charlottesville, VA: The

Michie Co.
Moffitt, Robert. 1983. An economic model of welfare stigma. American

Economic Review 73 (December): 1023-35.

Ross, Heather L. and Isabel V. Sawhill. Time of Transition, The Growth of

Families Headec by Women. Washington, D.C.: The Urban Institute, 1975.
Sorenson, Annemette and Maurice MacDonald. 1983. An analysis of child support

transfers. In Judith Cassetty, ed., The Parental Child Support

Obligation. Lexington, MA: 0. C. Heath and Co.
U.S. Bureau of the Census. 1981. Current Population Reports, Series P-23,

Special Studies No. 112, Child support and alimony: 1978. Washington,

D.C., September.
. 1982. Statistical Abstract of the United States: 1982-83

(103rd edition), Washington, D.C.
. 1983. Current Population Reports, Series P-23, No. 124, Child

support and alimony: 1981 (Advance Report). Washington, D.C, May,
U.S. Department of Health and Human Services, Office of Child Support
Enforcement, Annual Report, December 1983.

Table 1

Percentage of Women with Children Present from an Absent

Father Awarded Child Support by Race, Marital

Status and Year of Disruption

All Races

Black

Nonblack

All Women

Never -Married
Ever -Married

Currently Separated

Separation began:
1979-82

1975-78

1971-74

1970 or before

Ever-Divorced

Divorced in:
1979-82

1975-78

1971-74

1967-70

1963-66

1962 or before

60.1%
(7680)*

31.5%
(1894)

69.5%
(5786)

12.0
(1468)

12.0
(890)

11.9
(578)

71.5
(6212)

48.8
(1004)

75.9
(5208)

43.9
(1336)

33.6
(473)

49.5
(863)

50.7
(481)

48.0
(100)

51.4
(381)

48.9
(448)

38.5
(148)

54.0
(300)

29.7
(182)

24.0
(96)

36.0
(86)

30.7
(225)

24.0
(129)

39.6
(96)

79.1
(4876)

62.3
(531)

81.1
(4345)

81.3
(918)

71.6
(95)

82.4
(823)

82.3
(1693)

63.7
(157)

84.2
(1536)

80.8
(1098)

66.7
(120)

82.5

(978)

76.0
(649)

61.0
(82)

78.1
(567)

69.8
(298)

44.1
(34)

73.1
(264)

58.2
(220)

41.9
(43)

62.1
(177)

Source: Tabulations from the March/April Match Files of the 1979 and 1982
Current Population Surveys.

*Number of women.

Table 2

Effect of Selected Factors on the Probability that Child

Support is Awarded to Mothers, Age 18 or Over with

Children Present from an Absent Father

Inaependent
Variables

All
Mothers

Never-Married
Mothers

Ever-Married
Mothers

EDUC

.029
(7.80)

-.001
(0.15)

.027
(8.14)

COLLGRAD

-.069
(2.05)

-.041
(0.60)

-.067
(2.22)

NEAST

-.055
(2.58)

-.032
(1.07)

-.042
(2.16)

NCENTR

.044
(2.13)

-.024
(0.87)

.047
(2.48)

SOUTH

-.031
(1.64)

-.012
(0.42)

-.029
(1.71)

SMSA

.038
(2.21)

.012
(0.47)

.028
(1.77)

CC

-.029
(1.47)

.029
(1.15)

-.033
(1.87)

PATERNR

.037
(4.66)

.012
(1.07)

.035
(4.83)

AGE

.007
(1.62)

-.002
(0.32)

• • •

AGESQ

-.0001
(2.19)

.00002
(0.15)

AGESQ

-.000
(2.19)

AGEDIV

• • •

BLACK

-.211
(7.99)

SPANISH

-.120
(4.34)

YEAR82

-.007
(0.47)

BLYEAR82

.079
(2.43)

NEVMAR

-.732
(28.09)

-.016
(0.52)

-.106
(2.33)

.053
(1.87)

-.016
(0.44)

.001
(0.81)

-.138
(2.55)

-.075
(2.97)

Table 2 (continued)

Independent
Variables

All

Mothers

Never-Married
Mothers

Ever-Married
Mothers

REMAR

-.040
(2.26)

-.013
(0.81) ■

SEP

-.285
(12.31)

-.252
(11.62)

SEPGT2

-.103
(3.60)

-.059
(2.17)

TIME

• • •

.013
(5.45)

LAW

• • •

• • •

-.017
(3.32)

BLTIME

• • •

• • •

-.005
(1.09)

BLLAW

• • •

.025
(2.29)

Likelihood
Ratio Test

2755.45**

24.72*

958.68**

Sample Size

7680

1468

6094

Mean of Dependent
Variable (AWARDCS)

.601

.120

.717

♦Significant at the 5% level,
**Significant at the 1% level,

Table 3

Factors Determining Amount of Child Support Due

for Ever-Divorced and Currently Separated

Women Due Support in 1978 or 1981

Independent
Variaoles

(1)

(2)

EDUC

NEAST

NCENTR

SOUTH

SMSA

CC

PATERNR

AGEDIV

YEARSDIV

NEWDIV

REMAR

BLACK

SPANISH

YEAR82

DIVYRGE79

124.66
(8.92)

261.73
(2.79)

119.71
(1.42)

90.99
(1.10)

239.95
(3.36)

-53.61
(0.63)

494.71
(14.00)

18.08
(4.36)

-53.74
(6.45)

-459.89
(5.05)

8.14
(0.12)

-353.50
(3.27)

-136.90
(0.99)

-391.99
(6.42)

125.89
(8.99)

263.16
(2.80)

123.43
(1.47)

92.65
(1.12)

240.63
(3.37)

-53.71
(0.63)

492.22
(13.91)

18.04
(4.35)

-48.68
(5.38)

-490.91
(5.24)

15.42
(0.22)

-321.71
(2.60)

-129.25
(0.93)

-447.72
(6.16)

156.66
(1.54)

Table 3 (continued)

Independent

Variables (1) (2)

EDIVYRGE79 ... -130.18

(0.57)

CONSTANT -714.69 -757.06

(2.98) (3.14)

R2 .11 .11

Sample Size 3542 3542

Mean of Dependent

Variable (CHSUPDUE) $1832.00 $1832.00

Table A-l
Definition of Variables

AWARDCS = 1 if child support is awarded and 0 otherwise.

EDUC = number of years of school completed by the woman.

COLLGRAD = 1 if woman is a college graduate and 0 otherwise.

SPANISH = 1 if woman is of Spanish origin and 0 otherwise.

NEAST = 1 if woman lives in the Northeast and 0 otherwise.

NCENTR = 1 if woman lives in northcentral states and 0 otherwise.

SOUTH = 1 if woman lives in the South and 0 otherwise.

SMSA = 1 if woman lives within an SMSA and 0 otherwise.

-CC = 1 if woman lives within the central city of an SMSA and 0
otherwise.

PATERNR = number of children under 21 fathered or adopted by ex-husband who
are living with their mother (or from absent father for never-
married) .

REMAR = 1 for remarried mothers and 0 otherwise.

NEVMAR = 1 for never-married mothers and 0 otherwise.

AGEDIV = woman's age at divorce or separation.

SEP = 1 for currently separated women and 0 otherwise.

SEPGT2 = 1 for women currently separated more than two years.

3LACK = 1 if woman is black and 0 otherwise.

YEARSDIV = years since the divorce or separation.

NEWDIV = 1 if the divorce or separation occurred during the year prior to
the CPS survey.

AGE = woman's current age.

AGESQ = AGE squared.

CHSUPDUE = dollars of child support due in 1978 or 1981.

DIVYRGE79 = 1 if the aivorce or separation occurred in 1979 or later and 0
otherwise.

BDIVYRGE79 = BLACK*DIVYRGE79.

Table A-l (continued)

YEAR82 = 1 if the observation is from the 1982 March/April match file, and
0 otherwise.

BLYEAR82 = YEAR82*BLACK.

TIME = last two digits of divorce or separation year minus 60, or 0 if
year of marital disruption is before 1961.

LAW = last two digits of divorce or separation year minus 74, or 0 if
year of marital disruption is before 1975.

BLLAW = ei_ACK*LAW.

BLTIME = BLACK*TIME.

ECKMAN

^JDERY INC.

JUN95

-To.Plcas,? N. MANCHESTER
INDIANA 46962 '